🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Happy to share our work. We make VAE-style generative autoencoders within invertible models with exact likelihood. Sharp samples, much (much!) higher performance and more stable training than equivalent VAEs! Preprint: arxiv.org/abs/2205.09546 Code: github.com/gisilvs/AEF

Since good news never comes alone, I am happy to share that "Nyström Kernel Mean Embeddings", my first postdoc paper at MaLGa, with Antoine Chatalic, @AlessandroRudi and @lrntzrsc has been accepted at ICML2022!

Our new ICML paper shows how to build more interpretable Gaussian-Process Generalized Additive Models (GP-GAMs). With Xiaoyu Lu and Alexis Boukouvalas. Here’s our GP-GAM on the SUSY dataset.

New ICML paper! We (myself, Veronica Tozzo, John Higgins, and Rajesh Ranganath) open-source new deep learning architectures and a single-cell dataset for benchmarking predictive modeling on sets: arxiv.org/abs/2206.11925

Attending #ICML2022 this week! Let me know if you would like to meet up and talk about Gaussian processes, sequential generative modelling or anything 😃

Amazing postdoc opportunity with Seth Flaxman and group to work on computational statistics and machine learning for impactful problems!

Yesterday, Adil Salim and myself presented a tutorial on "Sampling as First-Order Optimization over a space of probability measures" slides are available here akorba.github.io/resources/Balt… and video icml.cc/virtual/2022/t… Hope you enjoy it!

At ICML or watching from home? Join us today in rooms 321-323 or online to talk about continuous time systems in ml! We have an amazing set of speakers and contributed talks! Check out our schedule here: icml.cc/virtual/2022/w…

Interested in modelling video data using fast Gaussian processes? Come check out our workshop paper "Markovian Gaussian Process Variational Autoencoders" today at the #ICML2022 "Continuous Time Perspectives in Machine Learning" workshop 😃 icml.cc/virtual/2022/w…

GPJax is now accepted into JOSS 🎉 If you like working with Gaussian processes and have wanted to try JAX, then consider GPJax! Code: github.com/thomaspinder/G… Docs: gpjax.readthedocs.io/en/latest/ Paper: joss.theoj.org/papers/10.2110…

"jaxtyping" is an amazing way to catch shape errors in your #JAX code. Just add type annotations!🔍 (GitHub: github.com/google/jaxtypi…) And as of the new version, it now supports the holy grail of shape-checking: symbolic expressions!🎉🎉 For example:

JMLR has recently launched a Special Issue on ML for addressing problems of climate change! We welcome all submissions which use machine learning to address problems of climate change, including mitigation, adaptation, and climate science. [1/4]

Time to start thinking about that AISTATS submission! :) Abstract deadline: Thu Oct 6 AoE Paper deadline: Thu Oct 13 AoE Conference will be hybrid, in-person location in the US.

My student Kayo Yin needs your help. Her visa has been unnecessarily delayed, which would prevent her from coming to UC Berkeley to start her studies. Despite bringing all required documents, the Department of State refused to process the visa and it could take months to re-process.

NEW: We're increasing our minimum student stipend by 10% from 1 October 2022. Full details: ukri.org/news/ukri-incr…

Towards Healing the Blindness of Score Matching. (arXiv:2209.07396v1 [stat.ML]) ift.tt/QDcJ9Bt

Yitang Zhang's long-awaited paper is out. Can't claim to understand it, but will steal the idea of substituting [current year] for [large constant I'm not able to optimize]. arxiv.org/abs/2211.02515

Can perturbations help improve goodness-of-fit tests based on Kernelized Stein Discrepancy (KSD), in particular against multi-modal alternatives? Checkout our new #ICML2023 paper: “Using Perturbation to Improve Goodness-of-Fit Tests based on KSD” arxiv.org/abs/2304.14762

New study published in nature Medicine estimates that in 2021, 2.9 million children (4.2%) in the United States had experienced the death of at least one parent or a grandparent caregiver responsible for most of the basic needs of the child 🧵1/3 imperial.ac.uk/news/259813/un…

RNN memory (HiPPO 🦛 style, predecessor to S4/Mamba 🐍) for posterior over functions When my awesome students told me you can build a memory for random functions that you don’t even observe I was like 🤯 Preliminary but exciting, feedback welcome 🤗